Scenario One
A parent wants to explore the types of exposure. The select a particular exposure (e.g., TV) might impact their child. They search for the exposure and are given an overview of the known outcome categories. Visualise info about the uncertainty for each effect. They filter the results for their child’s age. They click into an outcome category to get more information on the result (e.g., text summary, info about reason for the uncertainty).
Task One
Explore the exposures that there are evidence for.
Effect Counts
General Categories
Subcategories
Don’t think this works.
Effect Sizes
General Categories
I think this is my preferred option, although it does have some issues.
Problems
- The only way to reduce the data (that I could think of) is to take the mean, but that gives a pretty inaccurate view of some of the categories.
Subcategories
Don’t think this works.
#### Mixed General and Subcategories
This also works, especially if I could combine the subcategories to show the ‘range’ of effect sizes. Could not come up with a way to do that though.
Task Two
Explore an overview of the outcome categories for the selected exposure category.
(E.g., if they have chosen ‘video games’)
Using a linerange
Notes
- Still need a PCA or similar for study quality. This is just using sample size.
Using another heatmap
I think I prefer the linerange.
Task Three
Visualise the uncertainty for the selected exposure.
(Not sure how to do this one)
Task Four
Filter the results by child’s age
Idea one
This is just the same as Task Two. E.g.,:
Task Five
Click into an otcome category to get additional information on the result.
(Can’t really do this is ggplot)
Scenario Two
A domain expert wants to explore evidence for screen time in more detail. Use the system to get an overview of the evidence (exposure, outcome, effect size). They search for the exposure and/or the outcome, and optionally filter by age. They click into a relationship and can access the forest plot of the original studies, and the link to the original meta-analysis (and other, related studies, uncertainty). Three levels: evidence, uncertainty, meta-data.
Task Two
Explore the evidence for the selected exposure.
This is the same as Scenario One. To look at a difference, let’s imagine that they are instead searching by outcome.
(E.g., they choose ‘body composition’)
Idea One
Task Five
Explore the underlying data and meta-data
I’ve realised that I’ll need to relink the original study-level data back to this to show original forest plots.
Otherwise, one option would be to display a relationship and show the forest plots in the tooltip?